Optimal quantization and efficient cooperative spectrum sensing in cognitive radio networks

Abstract

Hard decision combination is bandwidth-efficient but unreliable while soft-decision combination provides reliability but at the cost of much bandwidth consumption. Reporting quantized information from CR users achieves a trade-off between hard and soft decision combination. In this paper optimal quantization scheme which quantizes the local information in a way that ensures maximum probability of detection while restraining probability of false alarm is proposed. The optimal scheme is based on energy detection and search iteratively for local quantization thresholds. Moreover Smith-Waterman algorithm (SWA), a string matching algorithm widely used in bioinformatics for aligning biological sequences, is used for comparing reports of all CR users to each other and computing similarity index for each CR user. Robust mean and robust standard deviation are calculated of the similarity indexes and a threshold is found. CR users who have similarity index below this threshold are excluded from global decision combination and their reports are discarded. The local decisions of rest of users are combined using modified rules of decision combination to take a global decision. The optimal quantization scheme is compared with other schemes. Simulation results show that the optimal scheme with quantization thresholds performs better than the other schemes.